EPJ Nuclear Sci. Technol. 5, 12 (2019)© U. Köse et al., published by EDP Sciences, 2019https://doi.org/10.1051/epjn/2019032

NuclearSciences& Technologies

Available online at:https://www.epj-n.org

REGULAR ARTICLE

Heat exchanger design studies for molten salt fast reactorUğur Köse1, Ufuk Koç1, Latife Berrin Erbay2, Erdem Öğüt3, and Hüseyin Ayhan1,*

1 FİGES Engineering, R&D Center, Nuclear Technology Department, 06690 Ankara, Turkey2 Eskişehir Osmangazi University, Mechanical Engineering Department, 26040 Eskişehir, Turkey3 FİGES Engineering, Teknopark-İstanbul, Additive Manufacturing Systems Department, 34906 İstanbul, Turkey

* e-mail: h

This is an O

Received: 15 February 2019 / Received in final form: 8 July 2019 / Accepted: 20 September 2019

Abstract. In this study, conceptual design for primary heat exchanger of the Molten Salt Fast Reactor is made.The design was carried out to remove the produced heat from the reactor developed under the SAMOFARproject. Nominal power of the reactor is 3 GWth and it has 16 heat exchangers. There are several requirementsrelated to the heat exchanger. To sustain the steady-state conditions, heat exchangers have to transfer the heatproduced in the core and it has to maintain the temperature drop as much as the temperature rise in the core dueto the fission. It should do it as fast as possible. It must also ensure that the fuel temperature does not reach thefreezing temperature to avoid solidification. In doing so, the fuel volume in the heat exchanger must not exceedthe specified limit. Design studies were carried out taking into account all requirements and final geometricconfigurations were determined. Plate type heat exchanger was adopted in this study. 3D CFD analyses wereperformed to investigate the thermal-hydraulic behavior of the system. Analyses were made by ANSYS-Fluentcommercial code. Results are in a good agreement with limitations and requirements specified for the reactordesigned under the SAMOFAR project.

1 Introduction

The power production from the thermal power plants ispossible through a thermodynamic cycle. The heatproduced in the core of a nuclear reactor by a fissionablefuel is transported into a coolant. Then, heat is transferredto the working fluid by using heat exchangers (HX). Thisthermodynamic cycle is mainly either Rankine or Braytoncycle. Molten Salt Reactors (MSRs) have gained impor-tance and different initiatives have been brought. Asustainable secure nuclear future based on ThoriumMolten-Salt Nuclear Energy Synergetics (THORIMS-NES)[1–5] and a conceptual design of a Stirling engine with theMSR reactor [6] are typical examples.

In MSRs, the heat transfer between the radioactiveliquid fuel salt and the secondary salt or between this saltand the conversion working fluid is ensured by heatexchangers. In the MSRs, one of the most importantequipment after the core is the heat exchanger. Therefore,the design of the heat exchangers forMSRs is a crucial task.

The number of heat exchangers varies depending on thedesign of the core, the number of loops, and the type of thepower cycle. The primary loop and the secondary loop heatexchangers, preheaters, steam generators, after heaters,condensers and others are all different types of heat

useyin.ayhan@figes.com.tr

pen Access article distributed under the terms of the Creative Comwhich permits unrestricted use, distribution, and reproduction

exchangers with certain vital functions in the MSR cycle.They are the main components of the plant due to not onlytheir functions and the numbers but also as their differentconditions and tradeoffs depending on the location andconnections in the main cycle.

The heat exchanger design is also strictly subjected tochange with the properties of fluids flowing through theheat exchanger. The type of the coolant and working fluidshould be determined in advance. In the MSRs, theprinciple liquid fuel is preferred as a molten salt consistingof eutectic salts and operating at temperatures of 600–800 °C including LiF-BeF2-ThF4 for the case of thoriumafter many researches. The problem is the thermo-physicalproperties of these eutectics. When LiF-BeF2-ThF4 isconsidered, it is found that two different mole percentagescan be used. In the first salt, the mole percentages of LiF-BeF2-ThF4 are 72.7–15.7–11.6 whereas in the second one,percentages will be 70.11–23.88–6.01 [7]. Such a smalldifference causes dramatic changes in the properties. Forexample, between 553 and 673 °C, the viscosity of the firstsalt varies between 14.1 and 7.74 cp. Whereas the viscosityof the second salt at the temperatures between 557 and653 °C varies from 12.59 to 7.30 cp [7].When the viscosity isconcerned, the 3rd and 4th degree polynomials exist forthese salts, respectively. This gives an idea how difficult isthe design of a heat exchanger system in terms of suchunique properties of salts. The properties of fluids thereforeindicate an important and serious step in the design study.

mons Attribution License (http://creativecommons.org/licenses/by/4.0),in any medium, provided the original work is properly cited.

2 U. Köse et al.: EPJ Nuclear Sci. Technol. 5, 12 (2019)

The main design purpose is to supply the exact heattransfer at a given time duration during operation. Thetotal heat transfer area between the fluids, the overall heattransfer coefficient and temperature differences of fluids inboth sides are the main parameters to be known on one sideand to be determined on the other side. They are mutuallydependent items.

For MSFR, that needs much larger heavy elementsinventory than thermal MSR, there is an incentive tominimize the fuel salt volume, a limited fraction of whichbeing in the heat exchangers. Therefore, the rate of heattransfer in the heat exchanger should be augmented andhence compact structures with high heat transfer areas perunit volume are necessary. However, enhancement in heattransfer will adversely affect pressure loss, which is relatedto pump design.

As it can be deduced from these explanations, thedesign of heat exchangers is critical. The design studies ofthe heat exchangers obviously require attention forcomplete and successful operation and it is needed thatdesign parameters must be strictly defined to get mostefficient heat transfer through the MSR. The previousefforts on the heat exchanger design for MSRs aresummarized first in the following section. Then the designstudies of the primary heat exchangers of Molten Salt FastReactor (MSFR) are carried out by FİGES under theSAMOFAR Project are going to be presented.

2 Historical development of the HXsfor MSRs

The number and size of heat exchangers in MSR plantsdepend on the physical properties of the fluids, power leveland the type of the thermodynamic cycle preferred. In theBrayton cycles, two or three to eight intercooling stages areused, whichmeans that there aremanyHXs as in themulti-reheat steam circulations in Rankine cycle.

The thermodynamics of MSRs is well established sincethe first studies carried by Oak Ridge National Laboratory(ORNL) in 1950s. The design experience on the HXs usedin MSRs lies to the initial development of MSRs to providea heat source to a jet engine in the US Nuclear AircraftDevelopment Program (NAP) in 1950s (or it is mentionedas Aircraft Nuclear Propulsion � ANP Program [8]).Following the cancellation of the aircraft program, MSRwas investigated between 1960s and 1970s up to thecancellation of the program by US ultimately. Due to over20-yr effort, HX design studies of ORNL deserves attentionbefore practicing the advanced design studies for MSFR.

The HXs which are used in the Homogeneous ReactorTest (HRE-2) at ORNL [9] were manufactured byconsidering typical shell & tube HX design principles.For the primary and secondary loops, eight fuel to sodiumHXs and sodium to steamHXswere all shell & tube type. Inorder to get an idea about the design, it is better to givesome numerical values for some parameters. The maincharacteristics of the HRE-2 steam generators had the heattransfer area of 44.6 m2 with the rate of heat transfer of5000 kW. The tube side diameter of 0.009525m was usedwith the velocities of 20.42 and 3.44m/s on the shell side

and tube side, respectively. The inlet and exit temperaturesfor shell and tube sides were 82.2–443.9 °C and 256.9–300.0 °C, respectively. The shell & tube steam generators inHRE-2 were thermal cycled with diphenyl as the heatingmedium. HRE-2 spare steam generator [9] containedeighty-eight 5/8 in OD, 0.095 in thick, type-347 stain-less-steel tubes having multiple U-bends.

A salt-to-gas primary HX design study was carried [10]for determining the problems and the effects of varying HXtubing size, coolant inlet temperature, coolant pressurelevel, allowable salt pressure drop and uranium enrichmentof the molten salt. For this design study, a reactor of640 MWth and electrical output of 275 MW wasconsidered. The type of the HX was a cross, countercurrentflow arrangement with molten salt having four serpentinepasses across the gas stream. Inconel was the material fortubing and circumferential fins.

Two experimental reactors were built and successfullyoperated. These were the Aircraft Reactor Experiment(ARE), the first MSR with 2.5 MWth, and the Molten SaltReactor Experiment (MSRE), with 8 MWth. MSRE was afirst experimental step to study a large Molten SaltBreeder Reactor (MSBR) and should have been followedby the Molten Salt Breeder Experiment (MSBE), a full-scale model of the MSBR used at 100–150 MWth [8,11]that has never been built. Some test loops with moltensalts were operated for hundreds of thousands of hours.Materials of construction were code qualified to 750 °C anda detailed conceptual design of a 1000 MWe MSBR wasdeveloped. The history of HXs for MSRs is parallel tothose efforts.

The HX in the primary fuel circuit of Molten SaltReactor Experiment (MSRE) of 10 MWth limited to about7.5 MWth was designed in 1961, fabricated in 1963 andinstalled in 1964. After some modifications, the HX hasbeen operated for approximately 14000 h with molten salttemperatures from 537.8 to 662.8 °C without any leakageand no change in the performance [12]. The HX of a shell &tube type with U-tube configuration was chosen since it isthe best design which satisfied the requirements andminimized the thermal expansion problems. The fuel flowsthrough the shell side of the primary HX and the coolantsalt circulates at the tube side. HX was tested with waterbut there were excessive vibrations and pressure drop. Itwas noted that [12] to alleviate the tube vibrations andlower the shell side pressure drop, 4 outermost U-tubes and4 associated tie bars were removed and plugs were weldedinto the 8 resulting tube stub ends, and into all the resultingholes in the baffle plates since the U-bends vibrated quiteseverely.

The heat-exchange system for one conceptual 1000MWeMSBR has been studied [13]. A modular-type design havingfour separatebut identical reactorswith their ownsalt circuitswas used by employing a two-region fluid-fuel concept inwhichfissilematerialswere in the core and fertilematerialwasin the blanket streams. Five types of HXs were mentioned ineach loop, namely, one primary fuel salt to coolant saltexchanger, oneblanket salt exchanger, fourboiler superheaterexchangers, two steam re-heater exchangers, and two reheatsteam preheaters to transfer the heat in the fuel and blanketsalts to the coolant salt and from the coolant salt to the

U. Köse et al.: EPJ Nuclear Sci. Technol. 5, 12 (2019) 3

supercritical fluid. Theywere all designed as shell & tubeHXsas two-pass vertical exchangers with disk and doughnutbaffles.

In the units for a 1000 MWe MSR plant [14], therewere a boiler superheater and two reheaters designed asshell & tube HX. Besides many different attempts todesign boilers for MSR plants with some seriousdisadvantages have been made, a reentry tube boilertype was proposed to satisfy all the major requirements forthe steam generators of liquid metal and MSR powerplants. The salt was NaBF4, which mainly determined thetube length required. It was added that plant layoutstudies in a conceptual design for the net electrical outputof 1000 MW favored the use of six steam generators eachcoupled to one of six fuel-to-NaBF4 HXs in the Rankinecycle. Therefore, the design was an illustrative for theanalysis of the full-scale steam generator. In this design,the thermal resistance of the inner tube wall at the lowerend was increased by using double-walled tube with a gapthat is going to be vented and filled with steam.

The steam generator with 349 parallel verticalU-shaped HX with one shell pass and one tube pass wassimulated [15] on an analog computer to understand thedynamic responses of the components. For the dynamicanalysis, a single water tubular channel surrounded by asalt annular channel was taken as a model. The designstudies of HXs should obviously include such dynamicanalysis.

A U-tube HX as one of the six fuel to inert salt HXs in a2200MWth reference design reactor was analyzed [16]. Thetube bundle had the tubes in an equilateral triangularlayout and the fuel salt flows axially around the tubes onthe shell side and the inert salt in counter flow inside thetubes. The analyses were concentrated on the parametriceffects to the HX design to optimize the system design. Inorder to minimize the fuel inventory in the HX, which is adependent variable of major interest, tube size is decreased,the fuel pressure drop is increased, it is decided that the saltcan be FLiNaK instead of NaBF4, and the temperaturedifference between the fuel and inert salt is increased.

InMSR plants, heat exchangers of the shell & tube typehave been distinguished so far. The well developedtechnology and widely used standards of the shell & tubetype HXs are the main reasons for their preferability.Mainly conventional shell & tube with U-tubes and shell &tube with U-tubes in a U-shaped casing for supercriticalpressure steam were proposed but the excessive tempera-ture difference between salt and steam, the possibility offreezing of salt and unstable boiling were major problems inthese types [14,17]. The temperature differences betweeninlet and outlet fluid during steady-state operation islimited but large temperature changes may occur attransient cases. Therefore, thermal stresses become amajordesign consideration too.

The main disadvantages of a shell and tube heatexchanger are low heat transfer efficiency, low heat transfersurface area density. Besides, the main advantages of platetype heat exchanger are less erosion-corrosion issuescompared to shell and tube heat exchanger, wide choiceof materials (important for corrosion and erosion of thesalts), low salt inventory, high turbulence and true counter

flow lead to an efficient heat transfer [18,19]. In this study,it has been focused on the plate-type heat exchanger designto overcome these problems.

By looking at these historical design efforts for thedesign of HXs for molten salt reactors, the HXs werespecified the existing technology and under the Standardsof Tubular ExchangerManufacturers Association (TEMA)and ASME. Today, technology presents well developedalternative compact types, like parallel plate HXs. It isobviously deduced that many critical parameters and therelations between these parameters must be strictly definedand considered to get an optimized system design as awhole. For the system designed, all boundary andoperation conditions were limited and studies were carriedout. All variables that affect the performance of the systemwere studied parametrically.

3 Design studies for SAMOFAR project

In this section, the design methodology and the proposedtype will be explained. The type of HX is determinedaccording to the design criteria for the adopted conceptualMSFR. The MSFR, which is designed under SAMOFARproject, has the thermal capacity of 3000 MW. There are16 HX located outside of the cylindrical reactor [20–23].Therefore, the thermal capacity of each heat exchangeris approximately 190MW. Since a compact heat exchangeris intended, it has been focused on plate type heatexchangers. Moreover, fuel residence time will be short,pressure dropamountwill be small and also themaintenanceprocedure is easier for the plate type HX. The design ofthe heat exchanger plates was carried out in a flat andcorrugated form. Analyses were conducted in such a waythat there were countercurrent flow for both designs. Sincethe circulation time of the fuel salt in the system shouldbe kept as short as possible [21,22], the single-passheat exchanger design, rather than the multi-pass, isconsidered.

3.1 Design procedure

In the heat exchanger design process, there are severalimportant requirements such as thermal-hydraulic design,mechanical design, material selection, manufacturability,maintenance and safety. The heat exchanger is expected tomeet certain criteria such as thermal capacity, hydraulicbehavior and material strength, especially, and to have themost suitable form such as compactness, manufacturabilityand maintainability. This study focuses on the thermal-hydraulic analyses of the heat exchanger. The designprocedure created for the study is given in Figure 1. One-way arrow represents forward process and two-way arrowrepresents iterative process. 3D CFD results were used tocorrelate the dimensionless model (0D Model). 3D CFDstudies and mechanical analyses provide lots of feedback toeach other.

In the flat plate design, parameters that affect theperformance of the system such as channel spacing, platethickness, fluid inlet velocities and inlet temperatures havebeen investigated by using a 0D Model with the solver

Fig. 1. Design procedure.

4 U. Köse et al.: EPJ Nuclear Sci. Technol. 5, 12 (2019)

created in MATLAB platform. For the decision ofgeometric configuration, 2 and 3 dimensional (2D&3D)computational fluid dynamics (CFD) analyses wereperformed. The flow distribution and thermal performanceof the system were examined in detail. In the 2D study,analyses were performed to ensure that the fluids wereuniformly and evenly distributed within the channel.Geometric parameters determined in 0D model were usedin 2D studies. In 2D analyses, collector locations, entrancegeometry type and flow separator position and dimensionswere studied. After achieving the ideal combinations, 3Dstudies were performed using 2D CFD results and 0DModel parameters like plate dimensions and inlet con-ditions. In 3D CFD studies, the thermal behavior of thesystem was investigated in detail. Bulk temperatures, localminimum temperatures and temperature gradient distri-butions were investigated.

In the corrugated plate design, the effect of designparameters such as groove depth, groove opening, channelwidth and fluid velocities were investigated by using 2DCFD analyses. Then the detailed 3D CFD analyses wereperformed to investigate the flow characteristics as well asthe thermal behavior.

3.2 Preliminary studies

For heat exchanger design, a program (script) was writtenin MATLAB and the user interface was prepared todetermine the geometric configuration. This programprovides a dimensional comprehension of the desiredlogarithmic temperature difference in a flat plate withoutseparators. The model was initially considered ideal. Forexample, collectors are properly positioned, flow isuniformly distributed and inlet effects are ignored. Withthe outputs of this program, the dimensions of the flat plate

were roughly decided and the CFD studies were continued.In line with the outputs from MATLAB, 3D unit-channelCFD analyses were performed on a flat plate with nobrackets by accepting that the flow is evenly distributed.The CFD results for bulk temperatures and system heightwere compared with the results of the MATLAB program.Considering the differences, the 0DModel was corrected byadjusting the correlations or revising the assumptions.After ensuring consistency between MATLAB code andCFD results, the channel parameters and boundaryconditions for a flat plate were studied on the code andthe operating range for an ideal heat exchanger wasdetermined.

After determining the working range for heat exchangerdesign, feasibility studies were performed on this design.The 2D CFD analyses were performed for suitable cases.Firstly, inlet and outlet collectors were positioned. Severalconfigurations were studied for an ideal location. Followingthis study, 2D and 3D CFD analyses were carried out byplacing separators in various numbers, thicknesses andpositions in the plate in order to ensure the reproducibilityand even flow distribution. After deciding the placement ofcollectors and separators, the velocity and pressuredistribution profiles of fluids in the plate were examined.In particular, the pressure distribution profiles giveinformation about which channel in the plate should benarrow and which should be wider. The heat exchangergeometry has been optimized by adjusting with theseparator locations and the channels. CFD studies wereperformed for several plate configuration to achieve besthydraulic performance. Velocity distributions of eachconfiguration are illustrated in Figure 2.

For this design study, stagnant flow or highly vortexflow conditions were avoided. In both conditions, uniformtemperature distribution will not occur and it may cause

Fig. 2. Velocity distribution of 2D parametric studies for flatplate (plate height is about 1.5m and width is 0.7m).

U. Köse et al.: EPJ Nuclear Sci. Technol. 5, 12 (2019) 5

local freezing. Besides, the uniformly distributed mass fluxis preferred in this study to control the temperature dropand to obtain a smaller temperature gradient in alldirections. As can be seen from Figure 2, there is auniformly distributed velocity profile along the channelcross-section in the final design.

In addition to the flat plate geometry, design studieswere carried out for the corrugated plate. 2D CFD analyseswere performed for corrugated plate. The distribution ofthe flow within the channels and the temperature behaviorwere investigated. Different combinations of channels havebeen studied and the effect over the flow distribution andtemperature profile is examined. Groove depth, grooveopening, channel width and fluid velocities were changedparametrically. Figure 3 shows the velocity distributionand temperature profiles of several geometric configura-tions. The whole system is shown to clearly see thetemperature change amounts, while a small part of thesystem is shown to see in detail the velocity behavior. In allcases, the mass flow rate is the same as each other.

As seen in the sub-figures, as grove depth (a) decreases,pressure drop amount decreases; however, temperaturedifference decreases too. In that case velocity magnitude isat low level within the channels. On the other hand, asgrove opening (b) decreases, heat transfer amountincreases; however, pressure drop amount increases too.It should be noted that, to reach the required temperaturedrop amount, the height of the HX will be different in allcases. So in the case of a smaller grove depth, the systemheight will increase and also pressure drop amount willincrease too. Besides, in the case of smaller grove opening,local freezing may occur due to the stagnant flow near theplate wall. However, this effect may prevent the plate fromthe salt corrosion.

After 2D studies, 3D CFD analyses were performed fora small part of this geometry. The curved structure of thisgeometric design made it very challenging to create themeshmodel. Velocity and temperature distribution profiles

were also investigated with 3D CFD analyses. For thecorrugated design, there is no spacer or supporter betweenthe plates. Since the grooves contact each other in theopposite direction, they serve as supporter at the sametime. However, stagnation regions will occur around thesecontact points. Due to the occurrence of stagnant regions(and also freezing regions), flat plate model is adopted asdesign geometry.

3.2.1 Zero-dimensional modeling

Heat exchanger geometry is divided horizontally by anumber of elements (nodes). The expected fuel tempera-ture drop (DTfuel) is distributed to the elements so thatthe temperature of each element will change as DTnode(DTfuel /# of elements). In other words, each horizontalelement is not equal in height, but DTnode between eachelement are the same. In this way, the value of the fueltemperature in each element can be clearly known whichled to easier calculations, however only the inlet nodetemperature of the coolant is known. The heat from eachelement of the fuel channel is transferred to the coolantchannel.

The energy that the element has is transferred viaconduction and convection heat transfer to the neighboringchannel. In order to achieve energy balance in the system,the heat transferred from the fuel channel, the heatreceived by the coolant channel, and the overall heattransfer must be equal to each other. Therefore:

_mfcp;fDTf ¼ _mccp;cDTc ¼ UATDTlm ¼ Q ð1Þequality must be provided for all neighboring nodes.WhereQ is the total heat rate, _m is the mass flow rate, cp is thespecific heat capacity, DT is the temperature difference,DTlm is the logarithmic mean temperature difference, AT isthe total heat transfer surface area, and U is the overallheat transfer coefficient. Lower indicesm, f and c representthe material, fuel and coolant salt, respectively.

The logarithmic mean temperature difference is definedas,

DT lm ¼ DT 1 � DT 2ð Þ=ln DT 1=DT 2ð Þ ð2Þwhere DT1=Tf,inlet�Tc,outlet and DT2=Tf,oulet�Tc,inlet.The overall heat transfer coefficient is calculated as,

1

U¼ 1

hfþ tmkm

þ 1

hcþRfm þRcm ð3Þ

where h represents the convection heat transfer coefficientand it can be calculated using several correlations such asthe original Dittus-Boelter correlation or Gnielinskicorrelation [24]. k, tm and R are the heat conductioncoefficient, material thickness and the thermal resistance,respectively. Lower indices fm and cm represent the fuel tomaterial interface and coolant salt to material interface,respectively.

So, using equation (1), the unknown temperature ofeach coolant nodes can be calculated with starting from thecoolant inlet. The element heights which can give the

Fig. 3. Velocity distribution and temperature profiles of 2D parametric studies for corrugated plate.

6 U. Köse et al.: EPJ Nuclear Sci. Technol. 5, 12 (2019)

desired temperature drop in the elements are alsocalculated one by one during the calculations. So the totalheight of the HX unit will be calculated.

Then, using the calculated values such as height,thermo-physical properties, and hydraulic properties of thenodes, the pressure drop (Dp) of the fuel and coolantchannels are calculated using,

Dp ¼ f � H

Dh� r � V

2

2ð4Þ

whereH,Dh, r andV represent height, hydraulic diameter,density and velocity, respectively. f represents the Darcyfriction factor and it can be calculated using the Swamee–Jain equation, which is the open form of the Colebrook-White correlation for the rectangular ducts as,

f ¼ 1:325 ln∈=Dh

3:7þ 5:74

Re0:9

� �� ��2

ð5Þ

where ∈ is the pipe surface roughness and Re is theReynolds number.

3.2.2 CFD modeling

The flow regime in both fuel and the salt channel isturbulence. So, in CFD analysis, the turbulence effect wastaken into account with the k-e (realizable)model. The walleffect is treated with Enhanced Wall Treatment option forsolid surfaces where the wall boundary condition is applied.

Viscous heat effects were also included in the analysis. The“Coupled” algorithm was used for pressure–velocitycoupling. The discretization methods were selected as“Presto” for pressure, “Quick” for energy and “SecondOrder” for other terms.

Fluid velocities and temperature conditions were usedfor inlet boundaries. Simulations were performed for oneplate and neighboring half cannels. Symmetry boundaryconditions were used for half channel interfaces. Residualsas a convergence criteria were adopted as 10�4 for allvariables.

In the 2D CFD analyses map mesh is used. The totalmesh number is about 1 million and y+ value is less than 2in this models. In 3D CFD models cubic (hex) meshelements are used. Edge-Sizing and Bias-Factor are definedfor each side of the middle portion of the lower channel-forming plates. The purpose of using these methods is toanalyze the thermal-hydraulic effects better by giving amore frequent cubic mesh to the transition regions betweenthe fuel, material and coolant. This is so important for theconjugated heat transfer calculation. In addition, the sweepmethod was used in the whole model in order to ensure thesufficient orthogonal quality and skewness values of themesh elements. A total of 15–20 million mesh elementswere used in 3D CFD calculations. Also y+ value is lessthan 2 in 3D studies.

Mesh sensitivity analyses was performed for 3D CFDcalculations. Some of the monitored values are listed inTable 1. The terms nfch, np and nax represent the number ofdivisions of flow channels (both for fuel and coolant), plate

Table 1. Monitored results for mesh sensitivity analyses.

nfch np nax Tfo (°C) Tco (°C) Tf,min (°C) DPf (bar) DPc (bar) Wch (MW)

16 10 200 664.6 598.5 615.8 2.66 2.13 0.5324 10 200 665.0 598.4 616.0 2.67 2.17 0.5332 10 200 665.0 598.4 616.0 2.67 2.17 0.5324 4 200 665.0 598.4 615.5 2.67 2.17 0.5324 8 200 665.0 598.4 615.9 2.67 2.17 0.5324 10 200 665.0 598.4 616.0 2.67 2.17 0.5324 10 50 664.9 598.4 614.8 2.67 2.18 0.5324 10 100 664.9 598.4 615.4 2.67 2.18 0.5324 10 200 665.0 598.4 616.0 2.67 2.17 0.5324 10 500 665.1 598.3 616.0 2.66 2.17 0.53

Fig. 4. Schematic view of the selected HX design with thecollectors (H=height, W=width, L= length of the HX).

Table 2. Boundary conditions of the HX.

Parameters Value

Fuel inlet temperature ∼750 °CCoolant inlet temperature ∼560 °CFuel inlet velocity ∼5m/sMaximum fuel velocity in the channel ∼2.5m/sCoolant inlet velocity ∼10m/sMaximum coolant velocity in the channel ∼5m/s

Table 3. Resulting values of the HX.

Parameters Value

Plate thickness 2 mmSeperator thickness 2 mmSeperator height ∼1.3 mFuel channel thickness 2 mmCoolant channel thickness 2–4 mmCollector diameter 0.21 mFuel outlet temperature ∼675 °CTemperature drop in HX ∼75 °CCoolant outlet temperature ∼620 °C

U. Köse et al.: EPJ Nuclear Sci. Technol. 5, 12 (2019) 7

and vertical direction, respectively. For all condition, outletbulk temperatures, local minimum fuel temperature,pressure drop amounts and heat transfer amount arechecked. According to the results of sensitivity analysis,the case ofnfch=24,np=10andnax=200was selected as themost suitable mesh structure of the geometry and used in allcases.

Minimum temperature at fuel channel ∼620 °CHX dimension (H�L�W) ∼2.0�1.5�0.7 mPressure drop in fuel channel ∼3 barPressure drop in coolant channel ∼5 barTotal fuel volume in 16 HX(HX and Collectors)

∼6 m3

3.3 Conceptual design for MSFR

According to the findings from CFD studies, the final formand geometric design is selected for HX of MSFR. The finalgeometric configuration of the plate type HX is shown inFigure 4. For the final design, 3D CFD analyses wereperformed with the boundary and operating conditionspresented in the EVOL project report [20]. The referenceconcept of MSFR is designed for a nominal power of3 GWth, with a salt temperature rise preliminary fixed atDT=100 °C [20,25]. Therefore, there must be a tempera-ture drop of 100 °C in the fuel salt channel of HX. It is alsostated that the fuel and coolant salt inlet temperaturesshould be around 800 and 490 °C, the fuel and coolant inletvelocities should be around 2–2.5m/s and 5–5.5m/s,respectively [25].

Many different geometric and thermal-hydraulic designstudies have been carried out in order to achieve 100 °Ctemperature drop with these boundary conditions. Thedegree to which these studies meet the design criteria isexamined.The design criteria are particularly focused on thecriteria such as the height and width of the heat exchanger,the volume of the fuel salt, the amount of heat transferred,minimumfuel temperature, temperaturedropof fuel saltandthe pressure drop for fuel and coolant channels.

Fig. 5. Velocity and temperature distributions in the fuel channel and the coolant channel (total height is 1.97m, plate width is 0.7m,there are 20 pieces equally spaced separator).

8 U. Köse et al.: EPJ Nuclear Sci. Technol. 5, 12 (2019)

In subsequent studies, it is recommended by the expertsthat the core outlet temperature will be below 750 °C forMSRs due to the salt corrosion over the material [26,27].Thus, the design criteria for heat exchanger design werefurther tightened, and studies continued for core outlettemperature (HX inlet fuel salt temperature) of around750 °C and temperature drop of 75 °C. The adoptedboundary conditions for thefinal design are given inTable 2.

Previous studies were repeated on these boundaryconditions. Both hydraulic and thermal design studies wereperformed for these conditions, and final geometric design isdetermined. Table 3 shows the geometric parameters andthe results obtained. The results provided all the require-ments. The temperature decrease amount in fuel side isobtained as 75 °Cwhich is adopted due to the design criteria.The minimum fuel salt temperature in the HXwill be about620 °C. The freezing temperature of the fuel salt is about565 °C. Thus, there will not occur freezing region in thesystem. The coolant outlet temperature is about 620 °C. So,the temperature gradient across the plate is about 120 °C.Pressure drop is between 3 and 5 bar in the system. Thedimensions of the heat exchanger giving these result arearound 2.0� 1.5� 0.7m (H�L�W). According to thissystem, the fuel volume in heat exchangers (16 heatexchanger) and collectors obtain around 6 m3.

Temperature and velocity distributions in the fuelchannel and the coolant channel of the HX is presented inFigure 5. As shown in the figure, uniform flow distributionoccurs in the system and temperature gradient across bothchannels is reasonable.

4 Conclusion

In this study, primary heat exchanger for MSR is designedtaking reference to the predefined operating and boundaryconditions. Since the compact HX is preferred, plate type

HX was selected. Both corrugated and flat plates wereinvestigated and analyzed. Due to the occurrence ofstagnation region in the HX, flat plate with separator wasused for the design.

Firstly, to obtain uniform flow distribution, size andlocation of the collectors and separators were determined.Then, boundary conditions and geometric dimensions wereoptimized to meet the thermal requirements. There areseveral significant criteria in the design. For example, thetemperature difference in the heat exchanger has to be75 °C, the maximum velocity in the channel has not toexceed the limits, the total fuel volume in the heatexchanger has to be below the limits and the minimumtemperature of the fuel salt will not lower than the freezingtemperature.

The temperature of the fuel salt is adopted as about750 °C at the inlet of HX. Coolant inlet temperature isadopted as about 560 °C. Fuel salt enters the HX from thetop side and exits from the bottom side. On the other side,coolant salt enters the HX from bottom side and exits fromthe top side. Velocity of the fuel salt is about 5m/s in thecollector region and 2.5m/s in the channel. Velocity ofthe coolant salt is about 10m/s in the collector region and5m/s in the channel. Salt molar content and thermo-physical properties of fuel (7LiF-ThF4-

233UF3) and coolant(LiF-NaF-KF) is adopted from reference study [25].Hastelloy-N is used for plate material [28].

According to the calculations, a compact HX isdesigned. The height of HX is about 2.0m, total lengthis about 1.5m and the width is 0.7m. Pressure drops forfuel and coolant channels were found as about 3 bar and5 bar, respectively. The total fuel volume will be about 6 m3

in the HX unit (in channels and collectors). The HXprovided the temperature drop of 75 °C in coolantside. The minimum fuel temperature is obtained about620 °C in the HX channels. The resident time is obtainedabout 2 s.

U. Köse et al.: EPJ Nuclear Sci. Technol. 5, 12 (2019) 9

The unreleased and commercial design data used in this studyhave been obtained through a delivery report of SAMOFARproject. Authors thank to the SAMOFAR project consortium forthe valuable supports.

Author contribution statement

Uğur Köse is a research and application engineer at FİGES.He is a staff of Nuclear Technology Department with aspecialization in nuclear thermal-hydraulics and nuclearsafety fields. He created the manufacturable CAD modelfor HX and he performed CFD analyses to optimize thethermal performance of HX. Ufuk Koç is a research andapplication Engineer at FİGES. He is a staff of NuclearTechnology Department with a specialization in nuclearthermal-hydraulics and code development. He developed0D based code to simulate HX system using MATLABtoolboxes and he performed CFD simulations to optimizethe hydraulic performance of HX. Latife Berrin Erbay is aprofessor at the Eskişehir Osmangazi University. She hasexperiences and publications of heat exchanger design andthorium molten salt reactors. She advised on the types andproperties of HX based on her background. Erdem Öğüt isthe Head of Additive Manufacturing Department atFİGES. He is an expert in mechanical properties of metals.He advised on manufacturability and strength of thedesigned geometry. Hüseyin Ayhan is aManager of NuclearTechnology Department at FİGES. He specializes innuclear reactor thermal-hydraulics, nuclear safety, ad-vanced nuclear reactor design, fluid mechanics and heattransfer applications. He guided the course of the design,taking into account the criteria and constraints.

References

1. K. Furukawa, L.B. Erbay, A. Aykol, A study on a symbioticthorium breeding fuel-cycle: THORIMS-NES through FUJI,Energy Convers. Manag. 63, 51 (2012)

2. K. Furukawa, K.Mitachi, S.E. Chigrinov, Y.Kato, A. Lecoco,L.B. Erbay, Rational Pu-disposition for 233 U-production byTHORIMS-NES (Thorium Molten-Salt Nuclear EnergySynergetics), International Atomic Energy Agency TechnicalDocument IAEA�TECDOC, 840, pp. 169–181, 1995

3. K. Furukawa et al., A road map for the realization of global-scale thorium breeding fuel cycle by single molten-fluorideflow, Energy Convers. Manag. 49, 1832 (2008)

4. K. Furukawa, L.B. Erbay, A study on a global scalesymbiotic thorium breeding fueal cycle, in Proceedings OfThe 2nd International Conference On Nuclear And Renev-able Recources NURER2010, 2010, pp. 255–261

5. K. Furukawa, D.E. Graves, L.B. Erbay, M. Hron, Y. Kato,New sustainable secure nuclear ındustry based on ThoriumMolten-Salt Nuclear Energy Synergetics (THORIMS-NES),Nuclear Power � Deployment, Operation and Sustainability(InTech Open, London, 2011)

6. L.B. Erbay, Examining the power generation by the stirlingheat engine combined with the Molten � Salt Reactor, inProceedings Of International Symposium On Efficiency,Costs, Optimization, Simulation And Environmental AspectsOf Energy Systems: ECOS’99, 1999, pp. 429–434

7. S. Cantor, Density and viscosity of several molten fluoridemixtures, ORNL-TM-4308, 1973

8. D. Scott, A.G. Grindell, Components and systems develop-ment for molten-salt breeder reactors, ORNL-TM-1855, 1967

9. J.A. Lane, H.G. Macpherson, F. Maslan, Fluid fuel reactors(Addison-Wesley Company, Inc., Boston, 1958)

10. R.E. Macpherson, Gas cooled molten salt heat exchanger �design study, ORNL-2605, 1958

11. C.W. Forsberg, Reactors with molten salts: options andmissions, The 2004 Frédéric JOLIOT&Otto HAHNSummerSchool � FJ/OH 2004, 2004

12. R.J. Kedl, C.K. McGlothlan, Tube vibration in MSREprimary heat exchanger, ORNL-TM-2098, 1968

13. C.E. Bettis et al., Design study of a heat-exchange system forone MSBR concept, ORNL-TM-1545, 1967

14. A.P. Fraas, A new approach to the design of design of steamgenerators for molten salt reactor power plants, ORNL-TM-2953, 1971

15. F.H. Clark, O.W. Burke, Dynamic analysis of a saltsupercritical water heat exchanger and trottle used withMSBR, ORNL-TM-2405, 1969

16. A.P. Fraas,M.E. LaVerne, Parametric survey of the effects ofmajor parametres on the design of fuel-to-ınert-salt heatexchangers for the MSBR, ORNL-TM-2952, 1971

17. L.B. Erbay, Knowhow Of Th-MSR heat exchange system,in 11th International Conference On Sustainable EnergyTechnologies, 2012

18. V. Ariu, Heat exchanger analysis for innovative moltensalt fast reactor, Master Thesis, Paul Scherrer Institute, 2014

19. Z. Pavel, V. Vaclav, Various methods to improve heat transferin exchangers, EPJ Web of Conferences 92, 02119 (2015)

20. Project No: 249696, Final Report � EVOL project, 201421. J. Serp et al., Themolten salt reactor (MSR) in generation IV:

overview and perspectives, Prog. Nucl. Energy 77, 308 (2014)22. M. Aufiero et al., Calculating the effective delayed neutron frac-

tion in the Molten Salt Fast Reactor: Analytical, deterministicand Monte Carlo approaches, Ann. Nucl. Energy 65, 78 (2014)

23. S. Wang et al., A passive decay heat removal system foremergency draining tanks of molten salt reactors, Nucl. Eng.Des. 341, 423 (2019)

24. F.P. Incropera, D.P. DeWitt, T.L. Bergman, A.S. Lavine,Fundamentals of Heat and Mass Transfer (John Wiley &Sons, Inc., 2007)

25. E. Merle-Lucotte et al., Preliminary design assessment of theMolten Salt Fast Reactor, in Proceedings of the ENC2012-Advanced Reactors, 2012, pp. 17–26

26. V. Ignatiev, A. Surenkov, Material performance in moltensalts (Elsevier Inc., Amsterdam, 2012)

27. V. Ignatiev, Materials and metals in MSR, in MSR SummerSchool, July 2–4, Lecco, Italy, 2017

28. Principle Features of Hastelloy-N. Available: https://www.haynesintl.com/alloys/alloy-portfolio_/Corrosion-resistant-Alloys/hastelloy-n-alloy

Cite this article as: Uğur Köse, Ufuk Koç, Latife Berrin Erbay, Erdem Öğüt, Hüseyin Ayhan, Heat exchanger design studies formolten salt fast reactor, EPJ Nuclear Sci. Technol. 5, 12 (2019)